Acoustic requirement of wind turbines can be efficiently meted using optimal design methods, which balance aerodynamic and acoustic goals. Yet, considering an additional discipline in optimization drastically increases the total cost. This paper proposes a cost-effective design optimization approach based on enhanced methods for both formulation and solution. Therefore, the problem is formulated using the MASSOUD deformation-based method, and then solved based on the cost-effective meta-models. The latter are constructed using an improved radial basis function (RBF) and an adaptive sampling, which is generated starting with a Latin hypercube database and enriched iteratively based on an exploration/exploitation criterion. The evaluation of the database is achieved by the validated aero-acoustics (CFD-CAA) simulations. The trained meta-model is coupled with the multi-objective algorithm NSGA-II in order to maximize the power coefficient and reduce the generated noise. The results show a maximum increase of 22% in the power coefficient at a tip speed ratio of 3.2. A consistent attenuation of higher-order tonal components is observed for all optimized configurations, with significant reductions at the second and third harmonics Moreover, the acoustic outcomes exhibit strong directional dependency. Noise reduction is observed in the upstream rotor region, accompanied by increased noise levels in the downstream zone. These findings are supported by detailed flow field analyses, which reveal the mechanisms governing aero-acoustics behavior.
Guermache et al. (Tue,) studied this question.